Heat Transfer Coefficients in the Boiling Section of a Long-Tube

true mean film coefficient of heat transfer. = thermal conductivity. = tube wall thickness, ft. = length of tube, ft. = constant in liquid temp. rise ...
0 downloads 0 Views 742KB Size
INDUSTRIAL AKD ENGINEERIXG CHEMISTRY

918

Nomenclature A U

b h IC

Lw

N S U

Literature Cited

= area of heating surface, sq. f t . = constant in Dittus and Boelter equation = constant in liquid temp. rise equation = true mean film coefficient of heat transfer = thermal conductivity = tube wall thickness, ft. = length of tube, ft.

= constant in liquid temp. rise equation

= = = = = = =

liquid velocity, ft./sec. heat trysferred, B. t. u./hr. temp., F. temp. drop across film Reynolds criterion, consistent units Prandtl criterion, consistent units Nusselt criterion, consistent units

Subscripts: A T

s B

L E

H

BPE F R

VOL. 29. NO. 8

apparent over-all temp. drop true over-all temp. drop = steam = vapor in evaporator = liquid at mean temp. = liquid at entrance temp. = temp. rise of liquid in tube = boiling point elevation = evaporator make-up feed = temperature loss from system = =

(1) Claasen, Mitt. Forschungsarb., 4, 49 (1902). (2) Cleve, Ibid., 322, 1 (1929). (3) Cryder and Gilliland, IND.EKG.CHEW,24, 1382 (1932). (4) Dittus and Boelter, Univ. Calif. Pub. Eng., 2, 443 (1930), Anal. Ed., 5, 359 (5) Hebbard and Badger, IND.ENG.CHDM., (1933).

(6) Jakob, Ann. Physik, 63 (6), 537-70 (1920); Badger and McCabe, “Elements of Chemical Engineering,” p. 603, New York, McGraw-Hill Book Co., 1931. (7) Jakob and Fritz, Forsch. Gebiete Ingenieurw., 2, 434 (1931). (8) Linden and Montillon, Trans. Am. Inst. Chem. Engrs., 24, 120 (1930).

(9) Logan, Fragen, and Badger, IND. ENG.CHEM., 26, 1044 (1934). (10) McAdams, W7.H., “Heat Transmission,” p. 161 et seq., New York, McGraw-Hill Book Co., 1933. (11) Meisenburg, Boarts, and Badger, Trans. Am. I r ~ s t . Chem. Engrs., 31, 622-38 (1935). (12) Nagle, Bays, Blenderman, and Drew, Ibid., 31, 593 (1935). (13) Nagle and Drew, Ibid., 30, 217 (1933). (14) Nusselt, 2. Ver. deut. Ing., 53, 1750, 1808 (1909); Mitt. Forschungsarb., 89, 1 (1910). (15) Schmidt, Schurig, and Sellschopp, Tech. Mech. Thermodynam., 1, 53 (1930). (16) Schmidt and Sellschopp, Forch. Gebiete Ingenieurzu., 3, 277 (1932). RECEIVED May 11, 1937. Presented before the meeting of the American Institute of Chemical Engineers, Toronto, Canada, May 26 t o 28, 1937.

Heat Transfer Coefficients in the Boiling Section of a Long-Tube, Natural-Circulation Evaporator‘ C . H.BROOKS2 AND W. L. BADGER University of Michigan, Ann Arbor, Mich.

T

Within the last few years in this country the LTV evaporator has come into some prominence for the handling of materials such as glue, sugar sirup, sulfate waste liquor, etc., of relatively small unit value but with high viscosities. The latter property makes them difficult t o handle in the standard types of natural-circulation evaporators, and the pumping cost is a disadvantage of the forced-circulation evaporator. It has been found that all of these materials, in addition to dilute salt brine, may be handled economically in this type of machine (26). Although many papers have been written about the Kestner evaporator, quantitative data on the subject are lacking. The only two papers that present experimental results (11, 16) report them in such a form that they cannot be interpreted. The other types of natural-circulation evaporators have Previous Work been more thoroughly investigated (4, 6, 10, 18, 19, 21). A number of papers in the foreign literature extol the virThe coefficients reported, either over-all or film, are based on tues of the long-tube, natural-circulation type; among its the entire length of the evaporator tube. They are usually advantages may be mentioned : spoken of as boiling coefficients but include the effect of a 1. Abilitv to handle heat-sensitive liquids without damage (I,certain but unkpown length of tube in which the liquid is not 16,98, 94, #7, R8). boiling. 2. Relative freedom from scaling trouble (3, %$,37, 88). This type of coefficient has been found to be a function of 3. Freedom from foaming - .(1. , 15,. 3.4, . . I T ). . the temperature difference (6, 19, WI), the liquid level (4, 6), 1 Complete tables of data obtained in these experiments will be published the temperature level (19, d l ) , and the concentration of the i n the Transactions oj the American Institute of Chemical Enyineers. solution (91). I n some cases the velocity of circulation was 2 Preaent address. Sun Oil Company, Marcus Hook, Pa.

HE long-tube, natural-circulation evaporator falls into the classification of the low-level, climbing-film machines. The most common representative is the socalled Kestner evaporator. This particular type was developed in Europe, beginning about 1900 (17). For some reason little interest was shown in this country until recently. Because of the redevelopment of this type of machine in America, entirely independent of the old Kestner evaporator, there is now an inclination to speak of it as the LTV (longtube vertical) evaporator. The characteristic feature of this type of evaporator is the length of the vertical tubes, up to 22 feet. In operation, the liquid is fed to the bottom of the tubes so slowly that it does not issue from the top as a solid stream but boils in the tube and is carried to the top as a spray or a climbing film.

INDUSTRIAL AND ENGINEERING CHEMISTRY

IUGUST, 1937

919

flowsheet of the equipment is determined (10, 19), but its given in Figure 1: effect on the coefficient could A method, as suggested by Boarts (81, not be determined since the has been developed for determining the Distilled water for feed wm r a t e of circulation was a drawn from storage tanks into length of the boiling section in an exthe feed weigh tanks, 1, from function of other 7-ariables. ternally heated tube. which the feed pump, 2, forced An exhaustive investigait into the double-pipe feed An empirical correlation of over-all cotion on an evaporator with a heater, 3. An e q u i v a l e n t single vertical tube is reefficients of heat transfer in the boiling secquantity of water was thus forced into the 2.002 inch ported b y K i r s c h b a u m , tion of a 2-inch 0 . d., No. 10 Birmingham 0.d. x 1.760 inch i. d. copper Kranz, and Starck (18). In wire gage, copper tube shows that the tube, 6, which was 20.0 feet this study the rate of circulalong between tube sheets. As coefficient is a function of the amount of tion, the liquid temperature the water flowed up the tube, it was heated by the condensavapor present, the weight rate of liquid gradient along and across the tion of steam in steam chest 7 tube, and the tube-wall temfed to the tube, and the average over-all until it began to boil. The perature gradient were detemperature difference between the vapor and liquid were separated by the deflector in vapor termined. A correlation of steam and the liquid. This correlation is head 9. The vapor was led their data is presented. It is applicable only within the ranges of varithrough the va or line, 16, to entirely empirical and is in condenser, 10, wtere it was wnables used for experimentation and to n o way general. d e n s e d . The unevaporated water, or to materials with similar physiwater left the vapor head Kirschbaum and hiL cothrough the thick liquor line, workers, however, did realize cal properties. 17, going to t)he thick liquor that “the way to the deThe mechanism of boiling in an extertanks, 11. The steam convelopment of the final equadensed in steam chest 7 flowed nally heated tube is exceedingly complithrough the condensate Ution probably leads to a divibend, 14, into s t e a m d r i p cated. Three types of action, froth, slug, sion of the heating surface tanks, 8. and film, have been found, but no definite into two parts, one for heatThe pressure in vapor head 9 ing the liquid and one for connection between the types of action was maintained c on s t a n t within =tl mm. of mercury by evaporation.” and the conditions of operation could be the manual and s o l e n o i d Considerable s t u d y h a s developed. vacuum breaks, 23, 22. The been made of the individual steam pressure was maintained coefficients of heat transfer within the same limits by means of the steam control between metal surfaces (invalves, 24, and the steam chest vent line, 15. ternally heated) and boiling liquids. I n general, the surface has been a flat plate in either a vertical or a horizontal posiMeasurements and Control Devices tion (9, 12, 13, 22). This work is of interest but as yet shows All temperatures were determined by means of copper-contoo great inconsistencies to be of any practical significance. stantan thermocouples. With the exception of the traveling thermocouple, the couples were inserted by sealing the wires, Experimental Apparatus with litharge and glycerol cement, into a 6-inch length of S/8-inch copper tubing which passed through a standard a/8-inch comThe investigation reported in this paper was carried outin pression fitting, drilled out with a No. W drill. a single-tube evaporator of the type known in commercial Couples of this type were used to determine the temperatures of the inlet to the evaporator tube, the unevaporated water outlet practice as the long-tube, natural-circulation evaporator. A Q

INDUSTRIAL AND ENGXNEERING CHEMISTRY

920

LEADS TO POTENTIOMETER 0.086"X 0.050'TUBE

I ,-UPPER CHAIN

MARKER

STUFFING BOX TO

CONDENSER

Y

TURNBUCKLE

VAPOR HEAD

S T E A M JACKET.

-16 STEEL SPRING WIRE-

PACKING GLAND FEED

T. C. -FROM

b

PUMP

FIG.,?. DETAIL OF TUBE CONNECTIONS

from the vapor head, the vapor in the vapor head, the steam enterin the steam chest, the steam in the steam chest (at the top, middfe, and bottom), and the condensate leaving the steam chest. These thermocouples were calibrated in an oil bath against a calibrated platinum resistance thermometer (6). The e. m. f. was read with a Leeds & Northrup portable precision potentiometer. The limit of error of the calibration was +0.04' F. The traveling couple (Figure 2) for the determination of the liquid temperature along the axis of the tube was of the same general type as that used previously (8,$0). No. 30 Birmingham wire gage, enameled, cotton-covered copper and constantan wires were inserted into an 0.086 inch 0.d. X 0.050 inch i. d. tube, 22 feet long. Provision was made for centering the lower end of the tube by vertical fins. The junction extended three-fourths inch beyond the lower end of the brass tube. The leads were withdrawn at the upper end. The traveling couple was calibrated by comparing the reading in its highest position with the temperature calculated from the vapor head manometer. Conditions in the evaporator were maintained such that fairly vigorous boiling occurred. It w a ~ found that the steam temperature could be varied within rather wide limits without appreciable change in the reading of the traveling couple. The limit of error of this calibration was * O . l O o F. The rate of feed to the evaporator was determined by means of the orifice installed between the control valves and the evaporator, as shown in Figure 1. The total quantity of feed delivered during a run was determined by difference at the feed weigh tanks. The steam chest and vapor head pressures were determined by means of mercury manometers. All manometer lines were of '/*-inch standard pipe.

VOL. 29, NO. 8

traverses of the tube with the traveling thermocouple were made. The values used for calculation were the average of the three readings in each case. The total heat input was used as the basis for calculation. Radiation was neglected. Slight superheat was maintained on the inlet steam. The increase in the sensible heat of the feed in being raised from the feed temperature to the maximum liquid temperature was subtracted from the total heat to give the heat transferred through the heating surface in the boiling section, (&/e),. This does not include the amount of heat produced by the decrease in the sensible heat content of the unevaporated liquid in cooling from the maximum liquid temperature to the vapor head temperature. The remaining items in the determination of the over-all heat transfer coefficient in the boiling section are fully discussed below. I n all investigations of this type the condition of the surfaces of the tube is important. An attempt was made in this work to duplicate commercial conditions as closely as possible. Approximately one hundred and fifty runs were made before the data used for this paper were taken. Steam from the University of Michigan powerhouse was used for the heating medium. Hence, the steam side must have closely duplicated the situation present in commercial machines. The condition of the liquid side was controlled by the following procedure: With a given set of feed rate, feed temperature, and boiling point conditions, three runs were made a t different values of the apparent temperature difference, spread over the entire range to be covered. Before check runs were made a t intermediate values of the temperature difference, the tube was boiled out with a dilute solution of inhibited hydrochloric acid. The results were required to be within *10 per cent of a smooth curve to be satisfactory. 3.6 3.4 32

30

3.6 I

$3.4

4

2 3.2 w

3.0 n

Experimental Procedure The following ranges of variables were studied: 4.1

Variable Feed rate to the tube lb./hr. Feed temp.. below vaGor headotemp., F. Boiling point in vapor head, F. Steam temp. minus vapor head temp., F.

Range 250 500 750 1000 10 55 d0 56, 60 156, l f 5 , 500 5-60

A run consisted of continuous operation under constant conditions for 20 minutes. All drip tanks and manometers were read a t regular intervals, the minimum number of readings being six. Difference readings on the tanks were determined during the run to ensure constancy of heat transfer rate. Three complete sets of thermocouple readings and three

3.9

3.7

B

2

4

6 8 IO I2 14 16 DISTANCE ABOVE LOWER TUBE SHEET, FEET.

I8

20

INDUSTRIAL AND ENGINEERING CHEMISTRY

AUGUST, 1937

Determination of Boiling Section

921

70 20

-

X As pointed out above, the liquid temperature distribution along the tube was deter10 mined during each run. S e v e r a l t y p i c a l E 8 Q I I_I curves for temperature vs. length are shown 6 c: in Figure 3. The outstanding characteristic 4 4 CY of all of these curves is the maximum which Lo is reached. 20 z W The length of the boiling section or the boilQ ing length, LE, was taken as the length between & = IO the point in the tube a t which themaximum E 8 temperature occurred, and the upper tube a 6 sheet. The choice of the maximum liquid tem3 perature as the division point between the seccr.4 m 20 40 80 100 200 400 20 tion of the tube used for heating and that in 40 60. 80 700 .. IOO .. - 400 .-_ m V, POUNDS OF VAPOR PRODUCED PER HOUR. 3 which boiling occurs, needs explanation. The liquid entering the tube, always beFIGURE 5 . U B vs. V low the boiling point in the vapor head during this investigation, will be heated as it flows up the The terminology developed may be more easily visualized tube. At the same time the pressure exerted by the material by Figure 4 where temperatures are plotted against tube farther up in the tube (both from static head and friction) length. The curve for liquid temperature is for the most will decrease. At some point in the tube, provided boiling general case-namely, a solution with an appreciable elevaoccurs, the temperature of the liquid will be such that its tion in boiling point and a feed temperature below the boiling vapor pressure will be equal to the pressure existing a t that point in the vapor head. The apparent temperature drop, At,,,., and the apparent temperature drop corrected for boiling point. If the liquid comes to equilibrium immediately, boiling point elevation, At,,,,., are shown as usually defined. The should occur when the vapor pressure of the liquid equals true mean temperature drop in the boiling section, At.,., is the static pressure. There is, however, some doubt that the mean height of the shaded portion. The true boiling coefequilibrium is reached a t once. Jakob and Fritz (14) and ficient, U,, reported in this paper is calculated on the surface Stewart and Hechler ($6)showed that it is possible to have corresponding to the length, LB, and on the mean temperasuperheating of the liquid before boiling starts. If this octure drop, Otav.. curs, the liquid will continue up the tube until some temperature is reached, a t which point vaporization will occur. Correlation of Results The vapor produced when the liquid boils flows up the tube UB, the average over-all heat transfer coefficient for the .to the vapor head. Hence, the pressure must decrease as boiling section, is mathematically defined as follows: the liquid progresses up the tube. No evidence has been advanced to indicate that superheating of the liquid is possible (&/Ob when boiling has once started. Hence, the temperature of lJB = the liquid in the boiling section must decrease as the distance It was believed that the most important variable affecting above the boiling level increases. the boiling heat transfer coefficient must be the velocity of Thus, the level a t which boiling starts is a t the point where the liquid. It was further believed that the controlling facthe maximum temperature occurs. The heat transfer area, tor in this velocity would be the quantity of vapor present in A B ,is the area of the evaporator tube above the boiling level. the boiling section. As the most practicable approach, the As shown in Figure 3, the temperature gradient of the total quantity of vapor was used. U B is plotted against V , liquid in the tube has no definite shape. Hence, the average the pounds of vapor produced per hour, in Figure 5. The average slope of the three groups containing the Iargest number of points was found to be -0.27. This average was determined by averaging the slope of the lines through the groups of points a t a constant feed temperature and feed rate, as well as the slopes of the bands. The remaining variables, W, (the feed rate to the tube), ALav., LE, and the temperature level were similarly investigated. The latter two had no appreciable effect. As shown in Figure 6, it was found that:

TUBE LENGTH

BOTTOM

DEFINITION

OF

PC

FIG 4. TERMS USED IN L.T.V. CALCULATIONS

temperature must be determined graphically. When the average liquid temperature is subtracted from the steam temperature, it permits the calculation of the average overall temperature difference in the boiling section, Atav..

m is shown graphically as a function of At.,. in Figure 7. I n 81 per cent of the 135 runs the calculated coefficient is within *20 per cent of the observed. It is to be emphasized that this correlation is entirely empirical. The information thus far obtained is not sufficient to permit the proposal of any general, theoretical equation for this type of heat transfer. Because of its empirical nature, this correlation should not be used for design purposes

VOL. 29, NO. 8

INDUSTRIAL AND ENGINEERING CHEMISTRY

922

IU

where the operating conditions are outside the range of variables u s e d during the experiments 1 00 u p o n w h i c h i t is based. 0 50 Owing to the nonuniformity of t h e l i q u i d temperature d i s t r i h u tion curves ( F i g u r e 3) 0 IO it is impossible to prea.05 5o d i c t t h e s e curves. I n 0 IO 20 30 40 At*" o r d e r t o p r e d i c t the value of Atav. from the steam temperature and the vapor head temperature, the curves in Figure 8 were developed. I n 85 per cent of the runs the calculated value is within *0.5" F. of the observed value. This is within the limits with which design data are usually known.

peared on the wall of the tube, above the point where the slugs broke. These experiments were repeated during this investigation, using a glass tube, approximately 3 / ~ inch 0.d., in a glass steam jacket. Barbet's results were verified; i. e., progressing up the tube the followingtypes of action were encountered: 1. A very short length in which the vapor and liquid were more or less uniformly distributed both transversely and longitudinally in the tube. 2. A section of slug action in which the transyerse distribution was uniform but the longitudinal was not; 1. e., there were alternate slugs of liquid and vapor. 3. A section in which film action occurred, or the longitudinal distribution was uniform, but the transverse was not; i. e., there was a core of va or with the liquid in a film of varying thickness on the walls oPthe tube. The annular rings mentioned by Barbet also appeared in this section. 5

2 s3 4

Discussion of Results This investigation is unique in that the heat transfer coefficients reported apply only to the boiling section. Previously reported results have combined the boiling and nonboiling (or heating) sections; the coefficients reported are a weighted average of the coefficients which apply over the two sections. The method used to determine the boundary between the two sections is that suggested by Boarts (8). However, his results of this type were too few to permit correlation. I n addition, they are completely out of the range of the present experiments. Inasmuch as this investigation was concerned primarily with the section in which boiling occurs, attempts were made to determine the phenomena in this section. The literature is deficient in this respect, yielding only two articles in which any attempt was made to explain the mechanism in the boiling section (7, 18). Barbet (7) describes in detail the appearance of the liquid when boiling occurs. He states that with low inlet velocities of the order of 0.3 foot per second, bubbles form which coalesce, pushing slugs of liquid ahead of them. Because of attrition a t the walls of the tube, these slugs become thinner and finally break, allowing the vapor to pass through. If vaporization is sufficient, the liquid on the walls of the tube moves upward. Mention was made that annular rings of liquid ap-

2

-

0

5

3 3

FEED RATE 500 L8lHR

2 100

500

1000

100

500

IO00

5000

5000

"LB

FIGURE 8. CORRELATION OF A(At) FOR VARIOUS FEEDRATES Correction to over-all temperature difference for the boiling section of 2-inch 0. d. tubes. A(A0 (Atover-all -At,,) O F. AtOver-dl (steam temperature - vapor head temperature) ' l?. Atav. = (steam temperature average liquid temperature) OF. LB =length of boiling section, feet. V = pounde of vapor per hour. P

AUGUST, 1937

INDUSTRIAL AND ENGIXEERING CHEMISTRY

I n the apparatus of Kirschbaum et al. ( I @ , a special glass plate was installed over the outlet of the evaporator tube. They observed themaction mentioned in the first two situations but not that in the third. Instead, the vapor and the liquid, in the form of drops, appeared to be uniformly distributed over the entire cross section of the tube. It is believed that a faulty method of observation caused Kirschbaum and his co-workers to draw erroneous conclusions concerning conditions in the tube. When a stream of liquid and vapor came out of the tube, some of it, a t least, must have impinged on the glass plate installed over the outYet of the tube. The vapor, acting on the film of liquid that would undoubtedly have been present on the glass, certainly would have caused ripples to form in the film. This would not allow the operator to make correct observations of the conditions a t the outlet of the tube. In the apparatus used for the investigation reported here, two sight glasses were installed a t the level of the upper tube sheet. These permitted observation of the outlet of the tube. With boiling lengths of one foot or less, froth appeared, corresponding to situation 1. With small temperature differences the liquid came out in slugs, traveling at high velocity with quiet intervals between the slugs. This corresponds to the second type of action. As the temperature difference and boiling length increased, two types of spray, coarse and fine, began to appear. The coarse spray seemed to be traveling a t a much lower velocity than the fine spray. I n addition, the fine spray seemed to be in the center of the tube, whereas the coarse was on the outside. Furthermore, the coarse spray appeared in most cases to be blown outward, but the fine spray traveled up to the deflector in straight lines. If there were a film on the walls of the tube, it would tend to be blown outward in the form of spray as soon as the retaining walls of the tube were removed. On the other hand, if there were drops of liquid in the vapor stream, they would tend to travel in a straight line for a time, owing to their momentum, even though the line of motion of the vapor changed. Hence, it, appears that action according to situation 3 does actually take place with the addition of a phenomenon that has not yet been explained. I n the study made by Kirschbaum and his associates (18), it was shown that in the boiling section the temperature distribution across the tube was practically uniform for almost t8heentire cross section, but that it climbed rapidly a t the walls. Since the static pressure a t the walls will be the same as that at the center of the tube, boiling should occur a t the wall. The formation of a bubble of vapor and its subsequent release into the main vapor stream bjt bursting, undoubtedly would tend to produce drops. Some of these drops might be thrown over to the other side of the tube-i. e., back into the liquid film. The remainder might stay in the vapor stream to form the fine spray observed. Hence, it would appear that in progressing upward from the boiling Ievel, a section of froth is found, then slug action, followed by film action, superimposed upon which there may be more or less spray. The temperature difference and the

923

boiling length, and possibly the amount of liquid present, determine the action occurring at the outlet of the tube. The relation between the operating variables and the types of action occurring is not yet known.

Nomenclature ( & / 0 )=~ heat transferred through heating surface in boiling section, B. t. u./hr. LB = length of boiling section, ft. = heat transfer area in boiling section, sq. ft. AB UB = av. over-all heat transfer coefficient in boiling section, B. t. u./hr./sq. ft./' F. = av. over-all temp. difference in boiling section, ' F. At,,. V = vapor produced, Ib./hr. = feed rate to tube, Ib./hr. Wf Acknowledgment The authors wish to express their gratitude to The Swenson Evaporator Company for their generous financial support of the research program upon which this paper is based and for permission to publish these results, and the following men for their assistance in operation and calculation during this investigation: W. C. Dresser, C. A. Framburg, Jr., L. H. Hilbert, R. H. Layer, J. R. Lienta, C. S. Moore, Jr., P. E. Negroni, K. E. H. Ness, M. H. Roth, J. F. Skelly, G. W. Stroebe, J. E. Thornton. Literature Cited Anonymous, Eng., 88,822 (1910). Aulard, Sucr. belge, 40,483 (1912). Badger, W. L., private communication. Badger, Trans. Am. Inst. Chem. I h g r s . , 13,Pt.11, 139 (1920). Badger and Hebbard, Ibid., 30, 194 (1934). Badger and Shephard, Ibid., 13,Pt. I, 101 (1920). Barbet, Bull. assoc. chim. sucr. dist., 32, 111 (1914). Boarts, R. M., Univ. Mich., doctorate thesis, 1937. Church and Cobb, S. M. thesis in chem. eng., Mass. Inst. Tech., 1922.

Cleve, Milt.Forschungsarb., 322, 1 (1929). Depasse, Bull. assoc. chim. sucr. dist., 37,434 (1920). Dunn and Vincent, S. M. thesis in chem. eng., Mass. Inst. Tech., 1931. Jakob and Fritz, Forsch. Gebiete Ingenieurw., 2,434 (1931). Ibid., 2,435-41 (1931). Janke, D. D. Zee, Bull. ussoc. chim. sucr. dist., 27, 616 (1909). Kerr, Trans. Am. SOC.Mech. Engrs., 38, 67 (1916). Kestner, Paul, British Patent 24,024 (1899), U. S. Patent 882,322 (1908); Dunn, J. E., Ibid., 917,258 (1910). Kirschbaum, Kranz, and Starck, Forsch. Gebiete Ingenieurw. Forschungsheft, 375, 1-8 (1935). Linden and Montillon, Trans. Am. Inst. Chem. Engrs., 24, 120 (1930).

Logan, Fragen, and Badger, ISD. ENQ.CHEM.,26, 1044 (1934). Moore, T$ans. Am. Inst. Chem. Engrs., 15,Pt. 11, 233 (1923). Neilon and McCormack, S.B. thesis in chem eng., Mass. Inst.

Tech., 1932. Reavell, J. SOC.Chem. Ind., 37, 172T (1918). Saillard, Sugar, 13,310 (1910). Stewart and Hechler, Refrig. Eno., 31, 107-11

(1936).

Swenson Evaporator Co., private communication. Ure, J. SOC.Chem. Ind., 43,294 (1924). Van Trooyen, Bull. assoc. chim.suci-. dist.,27,207 (1909). RECEIVED May 11, 1937. Presented before the meeting of the American Institute of Chemical Engineers, Toronto, Canada, May 26 t o 28,1937.